Inspection Device Using Fourier Transform for Signal Analysis

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

The randomness of the timing at which abnormal sounds are generated in time series data from sensors affects the accuracy of inspections using machine learning models, leading to insufficient inspection results in electric power steering devices.

Innovation Solution

The inspection device and learning model generation device employ short-time Fourier transforms to convert time series data into spectrogram data, and further Fourier transforms to exclude time components, allowing for determination of the inspection object's state based on frequency and amplitude components, thereby eliminating the influence of time component variability.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Ease of manufacture

If time series data is used directly for inspection, then the inspection process is simple, but the inspection accuracy deteriorates due to randomness in timing of abnormal sounds

Engineering Contradiction:

Improveinspection process simplicityVSAvoidinspection accuracy

Core Design Contradiction:

Ease of manufactureVSMeasurement precision

Solution Approach 1:

The patent transforms the data representation parameters by converting time series data into spectrogram data through short-time Fourier transform, and then into frequency-amplitude data. This parameter transformation eliminates the time component that causes randomness, thereby improving inspection accuracy while maintaining process simplicity.

Inventive Principle:

Principle #35Parameter changes

2Extent of automation

If machine learning is applied to time series data, then automatic inspection is achieved, but learning performance deteriorates due to time component variability

Engineering Contradiction:

Improveautomatic inspectionVSAvoidlearning performance

Core Design Contradiction:

Extent of automationVSReliability

Solution Approach 1:

The patent changes the data parameters from time-based representation to frequency-amplitude representation. This transformation removes the time component that causes variability in learning performance, enabling reliable automatic inspection through machine learning while maintaining automation.

Inventive Principle:

Principle #35Parameter changes

3Measurement precision

If Fourier transform processing is added to remove time components, then inspection accuracy improves, but processing complexity increases

Engineering Contradiction:

Improveinspection accuracyVSAvoidprocessing complexity

Core Design Contradiction:

Measurement precisionVSDevice complexity

Solution Approach 1:

The patent replaces direct time series analysis with Fourier transform-based frequency analysis. This substitution transforms the inspection mechanism from time-domain processing to frequency-domain processing, improving accuracy by eliminating time component variability while using standard signal processing techniques.

Inventive Principle:

Principle #28Mechanics substitution (Replace mechanical system)

Data Source

PatentUS11604170B2Inspection device and inspection learning model generation device

Publication Date: 2023.03.14 JTEKT CORP

AI summary

An inspection device includes a first data storage unit configured to store a first data which is time series according to a state of an inspection object, a second data generation unit configured to generate second data, which is a spectrogram including a first frequency component, a time component, and an amplitude component by performing short-time Fourier transform on the first data, a third data generation unit configured to generate third data including the first frequency component, a second frequency component, and the amplitude component by performing Fourier transform on time-amplitude data for each first frequency component in the second data, respectively, and a determination unit configured to determine the state of the inspection object based on the third data.